The economic value of ensemble forecasts as a tool for risk assessment: From days to decades

Abstract

SUMMARYDespite the revolutionary development of numerical weather and climate prediction (NWCP) in the second half of the last century, quantitative interaction between model developers and forecast customers has been rather limited. This is apparent in the diverse ways in which weather forecasts are assessed by these two groups: rootmean-square error of 500 hPa height on the one hand; pounds, euros or dollars saved on the other.These differences of approach are changing with the development of ensemble forecasting. Ensemble forecasts provide a qualitative tool for the assessment of weather and climate risk for a range of user applications, and on a range of time-scales, from days to decades. Examples of the commercial application of ensemble forecasting, from electricity generation, ship routeing, pollution modelling, weather-risk nance, disease prediction and crop yield modelling, are shown from all these time-scales.A generic user decision model is described that allows one to assess the potential economic value of numerical weather and climate forecasts for a range of customers. Using this, it is possible to relate analytically, potential economic value to conventional meteorological skill scores. A generalized meteorological measure of forecast skill is proposed which takes the distribution of customers into account. It is suggested that when customers' exposure to weather or climate risk can be quanti ed, such more generalized measures of skill should be used in assessing the performance of an operational NWCP system.

Seasonal predictability of daily rainfall statistics over Indramayu district, Indonesia

“…A set of retrospective seasonal forecasts from the ECHAM4.5 atmospheric GCM driven with constructed analog predictions of sea surface temperature (SST) were initialized on August 1 of each year 1979-2002(Li et al, 2007. In this "two-tier" system, SST is predicted on a monthly basis from the previous month (here July) using the constructed analog approach (van den Dool, 1994).…”

“…Palmer, 2002). Figure 9 shows the observed SAI time series for each rainfall statistic, together with box plots depicting the forecast ensembles.…”

Execution management in the GRID, for sensitivity studies of global climate simulations

“…However, the GRID cannot make the most of this kind of parallelism, since the latency across geographically distributed computers would render the program completely inefficient. Apart from computer parallelism, climate science is recently making use of a large number of simulations, referred to as an "ensemble", of the same phenomenon in order to assess the uncertainty inherent to the simulation (Hagedorn et al 2005;Palmer 2002). Ensembles of simulations with varying parameters are also used for sensitivity experiments and many other applications.…”

Developments in dynamical seasonal forecasting relevant to agricultural management

“…A general methodology for assessing the value of ensemble climate forecasts in end-user applications was discussed in Morse et al (2005). In particular, if users have quantitative application models requiring weather information as input, as in the case of crop models, these models can be directly linked to the output of individual members of the forecast ensemble (Palmer 2002). The net result is a probability forecast, not of weather or climate, but of a variable directly relevant to the user, e.g.…”